What problem does it solve? Knowledge scattered across chat sessions, local notes, GitHub issues, and memory files becomes impossible to find and maintain. This Skill provides a structured workflow for classifying, deduplicating, storing, and syncing knowledge across multiple storage layers so information lands in the right place every time. ## Core Features & Use Cases - Multi-layer storage routing: Classifies incoming knowledge and routes it to the right layer — Claude Code memory files, MCP memory knowledge graph, GitHub/Linear, a knowledge base repo, or external data stores like Supabase. - Deduplication-first ingestion: Searches existing memory files and MCP memory nodes before writing, updating existing entries instead of creating duplicates. - Sync operations: Syncs conversation sessions, workspace state, and cross-source exports (Claude, ChatGPT, bookmarks, GitHub activity) into a durable knowledge base repo with indexes and redacted secrets. - Use Case: After a long planning session, ask the agent to save the key decisions — it will classify each item, check for existing entries, write memory files with YAML frontmatter, create MCP memory entities and relations, and commit durable notes to the knowledge base repo. ## Quick Start Ask the agent to save this conversation's key decisions to the knowledge base and sync them across memory layers.
